Speizer et al. BMC Pregnancy and Childbirth 2014, 14:398 http://www.biomedcentral.com/1471-2393/14/398

RESEARCH ARTICLE

Open Access

Factors associated with institutional delivery in Ghana: the role of decision-making autonomy and community norms Ilene S Speizer1,2*, William T Story1 and Kavita Singh1,2

Abstract Background: In Ghana, the site of this study, the maternal mortality ratio and under-five mortality rate remain high indicating the need to focus on maternal and child health programming. Ghana has high use of antenatal care (95%) but sub-optimum levels of institutional delivery (about 57%). Numerous barriers to institutional delivery exist including financial, physical, cognitive, organizational, and psychological and social. This study examines the psychological and social barriers to institutional delivery, namely women’s decision-making autonomy and their perceptions about social support for institutional delivery in their community. Methods: This study uses cross-sectional data collected for the evaluation of the Maternal and Newborn Referrals Project of Project Fives Alive in Northern and Central districts of Ghana. In 2012 and 2013, a total of 2,527 women aged 15 to 49 were surveyed at baseline and midterm (half in 2012 and half in 2013). The analysis sample of 1,606 includes all women who had a birth three years prior to the survey date and who had no missing data. To determine the relationship between institutional delivery and the two key social barriers—women’s decision-making autonomy and community perceptions of institutional delivery—we used multi-level logistic regression models, including cross-level interactions between community-level attitudes and individual-level autonomy. All analyses control for the clustered survey design by including robust standard errors in Stata 13 statistical software. Results: The findings show that women who are more autonomous and who perceive positive attitudes toward facility delivery (among women, men and mothers-in-law) were more likely to deliver in a facility. Moreover, the interactions between autonomy and community-level perceptions of institutional delivery among men and mothers-in-law were significant, such that the effect of decision-making autonomy is more important for women who live in communities that are less supportive of institutional delivery compared to communities that are more supportive. Conclusions: This study builds upon prior work by using indicators that provide a more direct assessment of perceived community norms and women’s decision-making autonomy. The findings lead to programmatic recommendations that go beyond individuals and engaging the broader network of people (husbands and mothers-in-law) that influence delivery behaviors. Keywords: Institutional delivery, Ghana, Maternal health, Autonomy, Social support

* Correspondence: [email protected] 1 Carolina Population Center, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA 2 Department of Maternal and Child Health, Gillings School of Global Public Health, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA © 2014 Speizer et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Speizer et al. BMC Pregnancy and Childbirth 2014, 14:398 http://www.biomedcentral.com/1471-2393/14/398

Background As we approach 2015 and the deadline for attainment of the Millennium Development Goals (MDG), increased attention and effort are being given to reaching target populations with specific programmatic strategies, especially in countries that are not on track to attain the MDGs. In Ghana, the site of this study, where the maternal mortality ratio (MMR) remains high at 350 maternal deaths per 100,000 live births [1] and under-five mortality is estimated at 82 deaths per 1000 live births [2], there has been increased attention to initiatives to improve maternal, infant, and child health services. To help in the attainment of improved infant and child health (MDG 4) and improved maternal health (MDG 5), programs in Ghana and elsewhere promote skilled attendance at delivery; in many low income countries, this is equated with institutional (also called facility) delivery [3]. Skilled attendance at delivery means having an accredited health professional, including a midwife, doctor, or nurse, who has been trained in the skills needed to manage a normal or uncomplicated pregnancy and childbirth and to support the woman in the immediate postpartum period. This person should also be able to identify, manage and refer complications experienced by the woman or the newborn [4]. Numerous studies have demonstrated that in subSaharan Africa, including Ghana, there is often high use of antenatal care services but lower use of institutional delivery [2,5,6]. For example, the 2008 Ghana Demographic and Health Survey (GDHS) showed that more than 95% of women who had a birth in the last five years received antenatal care from a skilled provider prior to birth [6]. As compared to high antenatal care use, only 57% of women had an institutional delivery and only 59% delivered with a skilled attendant present [6]; similar distinctions are found in the 2011 Multiple Indicator Cluster Survey [2]. Wide variability in institutional delivery and skilled attendance at delivery was observed by region of residence with the Northern region having the lowest percentage of women delivering in a facility and the Central region falling in the middle on percentage of births in a facility [6]; these are the two regions covered in this study. A number of studies using the GDHS demonstrated important demographic- and policy-level factors associated with institutional delivery [7-9]. Recent global studies have examined common barriers to antenatal care and institutional delivery [3,10-17]. Much of this research has focused on transportation, distance, and cost [13,15,16]. A recent qualitative study by Matsuoka and colleagues in Cambodia (2010) demonstrated five types of barriers to utilization of government maternal health services [17]. These barriers were: financial; physical; cognitive; organizational; and psychological and social.

Page 2 of 13

Often, the financial and physical barriers are examined together to capture issues around transportation and distance to a facility as well as costs to reach or use the facility [13,15-17]. Ghana has implemented the National Health Insurance Scheme in an effort to reduce these types of financial barriers. Recent studies from Ghana have found that women with health insurance were more likely to have an institutional delivery [18-20] and insurance was associated with better maternal and child health outcomes [20]. Cognitive barriers are related to misconceptions about services offered and concerns about quality of services [17]. Organizational barriers are focused on the role of the providers in terms of attitudes, availability, and services offered [17,21]; these have been found to be particularly important from qualitative studies in Ghana [14,21]. Finally, psychological and social barriers are related to community norms and attitudes toward facility delivery and toward the staff at the facility [3,17]. These barriers to health care have been demonstrated in various cultural contexts including Bangladesh [22], Cambodia [17], Nepal [23], rural Kenya [13], and Northern Ghana [10] mostly using qualitative data or nationally representative data from Demographic and Health Surveys. Social barriers to institutional delivery, such as community attitudes towards institutional delivery and levels of decision-making autonomy among women, have received much less attention in the literature [24]. Community beliefs and attitudes about maternal health behaviors have been shown to influence a woman’s individual decision to seek care. For example, in a study of six countries in sub-Saharan Africa, Stephenson and colleagues [25] found that community norms about facilitybased delivery and women’s decision-making autonomy were potential pathways that influenced the decision to deliver a child in a health facility. In rural Tanzania, community beliefs that facility delivery is important for the health of the mother and baby were associated with use of facility-based delivery [26]. In a separate study among the same population in Tanzania, male partners’ opinions about institutional delivery were associated with actual institutional delivery, such that spouses who disagreed about the importance of institutional delivery were less likely to have one compared to spouses who agreed that delivering in a health facility was important [27]. In rural Mali, mothers-in-law’s beliefs and attitudes were demonstrated to have an influence on their daughters-in-law’s maternal health care-seeking behaviors [28]. Previous studies have also shown that women’s decision-making autonomy is associated with the use of health facilities for delivery. In Nigeria, women with greater decision-making autonomy were more likely to deliver in a health facility, which may indicate that these women were better able to advocate for and access a

Speizer et al. BMC Pregnancy and Childbirth 2014, 14:398 http://www.biomedcentral.com/1471-2393/14/398

health facility for childbirth [29]. In Bangladesh, households in which husbands made decisions alone were associated with less use of antenatal care and skilled delivery care compared to households that practiced joint decisionmaking [30]. The relationship between women’s decisionmaking autonomy and use of maternal health services may be due to women’s power to realize their preferences, which includes a stronger preference for ensuring their own health [31]. Although some studies have demonstrated the importance of community norms and women’s decision-making autonomy on the decision to deliver in a health facility, there have been few studies that have looked at the two pathways together and examined the ways in which community norms and household decision-making autonomy interact with one another. This study contributes to our understanding of autonomy and social barriers to institutional delivery in Ghana using recently collected quantitative data from two regions of Ghana. Because the focus of the study was to obtain information on barriers to institutional delivery and women’s referral experiences, this study includes a large sample of women who recently delivered a child, providing rich information on barriers to institutional delivery in these regions. The objectives of this study are to examine whether women’s decision-making roles and their perceptions about social support for facility delivery, measured at the individual and community levels, are associated with women’s actual place of delivery in Ghana. Furthermore, we examine how the relationship between community-level attitudes and institutional delivery differs for households in which women have a say in their own health care and those that do not.

Methods Data collection

The cross-sectional data for this study come from baseline and midline surveys that were used during the evaluation of the Maternal and Newborn Referrals Project of Project Fives Alive in the Northern and Central Regions of Ghana. The Maternal and Newborn Referrals project is being implemented by the Institute for Healthcare Improvement (IHI), the National Catholic Health Service (NCHS) and the Ghana Health Service (GHS). The evaluation is being led by the University of North Carolina at Chapel Hill and the University of Ghana. Baseline data were collected between May and June 2012 to help design the project and midline data were collected between October and November 2013 to help strengthen project implementation. Since the focus of this study is not significantly affected by the interventions implemented as part of the Maternal and Newborn Referrals Project and there were delays in project initiation until July, 2013, the baseline and midline data were merged to provide a larger cross-sectional sample.

Page 3 of 13

Multiple survey instruments were used at baseline and midline, including a household survey, a community leader survey, and facility surveys (with clients, providers, traditional birth attendants, and chemical sellers). This analysis focuses on the household survey data. The purpose of the household survey was to obtain information on knowledge, attitudes and practices regarding maternal and child health services. The household survey used the 30-by-N cluster sample design; this method is commonly used in child survival programs [25,32]. Cluster sampling is an efficient sampling method because it provides a means to obtain a representative sample from the region without undertaking a census of households in the community. However, cluster sampling leads to biased standard errors due to the correlation between observations from the same cluster. We explain our approach for accounting for the biased standard errors in the analysis section. The overall sampling strategy was designed to meet the evaluation objectives for the Maternal and Newborn Referrals Project [18]. At baseline, the goal was to include a large sample of recently pregnant women (pregnant in the last 12 months) to identify their experiences with pregnancy, childbirth, and newborn health. Thus, we used a 30-by-7 sampling approach to identify thirty clusters per region (thirty from the three districts in the Northern region and thirty from the three districts in the Central region), and seven recently pregnant women in each cluster were randomly selected for interview. Random selection of clusters was undertaken from an exhaustive list of communities in the six study districts. The recently pregnant women were randomly sampled from a list of all recently pregnant women in the community (determined through interviews with community leaders and health workers in the community). To supplement the sample of 210 recently pregnant women per region, we also included 14 nearby neighbor women (ages 15–49) who may or may not have been recently pregnant to permit an examination of maternal and newborn health knowledge, attitudes, and behaviors of women in the community. At midline, the same 30-byN cluster design was employed, however, a new sample of communities was drawn from the same districts. As with the baseline survey, in all selected clusters, seven recently pregnant women were surveyed as well as 14 nearby neighbors. For the purpose of this study, which uses baseline and midline data, and accounting for plausible design effect, our sample size is adequate to obtain precise estimates of our key outcome (institutional delivery). A total of 2,527 women were interviewed in the two rounds of data collection (1,267 women were interviewed at baseline and a new sample of 1,260 women were interviewed at midline). This analysis of institutional delivery focuses

Speizer et al. BMC Pregnancy and Childbirth 2014, 14:398 http://www.biomedcentral.com/1471-2393/14/398

on a sub-sample of the 2,527 women interviewed, which excludes women who did not have a birth in the last three years, were not in union, or had missing information on the key variables of interest. Thus the final analysis sample is 1,606 women. Ethics review approval for the study was obtained by the University of North Carolina at Chapel Hill and the Ghana Health Service. Informed consent was obtained from all study participants. Variables

The key dependent variable for this analysis is the place of delivery of the last birth in the last three years. Women who delivered in a health facility are coded one, whereas all women who delivered at home or in the home of someone else (e.g., a relative or a health worker) are coded zero (i.e., non-institutional delivery). The main independent variables for this analysis focus on decision-making autonomy and attitudes toward institutional delivery. First, all women were asked: “Who usually makes decisions about health care for you?” Women who reported that they make the decisions alone or make the decisions jointly with their partner were the reference group (high decision-making autonomy), and were compared to women who reported that their partner makes the decision alone (low decisionmaking autonomy). A third category was also created for the small number of women who reported that someone else makes the decision. The other independent variables of interest are attitudes toward institutional delivery, represented by three separate questions. First, all women were asked: “How many women do you think in your community deliver their baby in a health facility?” Response options were: none, some, most, and all (coded 1–4); the small number of women who reported “don’t know” (n = 114) were dropped from the analysis. Second, women were asked: “In your opinion, what percentage of men in your community is supportive of facility delivery?” Response options were: no men, few men, some men, most men, and all men (coded 1–5); the 149 women who reported “don’t know” were dropped from the analysis. Finally women were asked: “In your opinion, what percentage of mothers-in-law in your community is supportive of facility delivery?” Response options were: no mothers-in-law, some mothers-in-law, most mothers-in-law, and all mothers-in-law (coded 1–4); the 172 women who reported “don’t know” were dropped from the analysis. To examine community-level attitudes toward facility delivery, we also created comparable variables at the community-level for each of the three attitude questions. In particular, for each woman, we calculated the average response on how many women in the community she perceived had delivered their baby in a health facility.

Page 4 of 13

These community-level responses were calculated by creating an average value of all women living in the cluster, removing each individual woman from the calculation. A similar approach was undertaken for the community-level men’s attitude and the community-level mother-in-law’s attitude. All models control for demographic factors previously found to be associated with facility delivery including age, education, ethnicity, employment status, religion, parity, wealth, region, and time period (baseline or midline) [24]. See Table 1 for a description of these variables. The wealth variable was created based on three household characteristics: type of toilet, type of fuel used in the household, and location of the kitchen. Households with a non-improved toilet facility (as defined in the Ghana DHS), that use wood for their source of fuel, and that have a kitchen outside their household were coded as being the poorest households. Households with two out of three of these lower quality scenarios were considered to be medium, and households with none or just one of these lower quality scenarios were considered to be the richest. This is the same approach that was used in an earlier analysis of health insurance effects on facility delivery using these same data [18]. In the full sample, based on this classification, about 40% of the women were in the poorest category, 40% in the medium category, and only 19% were in the richest category (see Table 1). It is worth noting that use of antenatal care (ANC) during the pregnancy was not included as an independent variable in the reduced form models presented. Previous research has demonstrated that use of ANC is endogenous and would introduce bias into the models presented [33]. Analysis

We use bivariate analyses to examine the association between the key independent variables and institutional delivery, controlling for key demographic variables described earlier and adjusting for the clustered survey design. Because the dependent variable of interest (institutional delivery) is binary, and we are interested in both individual and community-level attitudes, we use multi-level logistic regression models. To examine if the relationship between community-level attitudes and institutional delivery differs by women’s decision-making autonomy, we use models with cross-level interactions between community-level attitudes and individual-level autonomy. Since the interactions cannot be evaluated by looking at the sign, magnitude, or statistical significance of the odds ratio for nonlinear models [34], we plot the interaction using the margins command in Stata 13. All regression analyses adjust for the clustered survey design by including robust standard errors in Stata 13 statistical software. Regression results are presented by showing the odds ratios and the 95% confidence intervals.

Speizer et al. BMC Pregnancy and Childbirth 2014, 14:398 http://www.biomedcentral.com/1471-2393/14/398

Page 5 of 13

Table 1 Characteristics of total sample (baseline and midline), recent birth sample, and analysis sample from Ghana evaluation of Maternal and Newborn Referrals Project, 2012, 2013 Characteristics

Full sample

Recent birth sample (birth in the last 3 years)

Analysis sample (in union, birth in the last 3 years, no missing information)

Percent Number (n = 2527*) Percent

Number (n = 1840*)

Percent

Number (n = 1606)

Factors associated with institutional delivery in Ghana: the role of decision-making autonomy and community norms.

In Ghana, the site of this study, the maternal mortality ratio and under-five mortality rate remain high indicating the need to focus on maternal and ...
299KB Sizes 0 Downloads 8 Views